1. Field of the Invention
This invention is generally concerned with the field of optoelectronic systems and devices. More specifically, embodiments of the present invention relate to a transistor header that includes various features directed to the enhancement of the reliability and performance of various electronic devices, such as lasers, included in the transistor header.
2. Related Technology
Fiber-optic and opto-electronics have become important components in modern networking circuits. Using fiber-optic circuits allows for efficient, accurate and quick transmission of data between various components in a network system.
As with the design of most any system, there are engineering tradeoffs that often have to be made when implementing fiber optic systems. For example, the size and modularity of components must often be balanced against the need for additional space to accommodate heat dissipation and circuit monitoring components. While it is desirable to minimize a component's size, some configurations have previously limited this minimization due to their inherent characteristics. For example, previously many lasers used in fiber-optic systems that have the characteristics needed for long-distance transmission and/or dense wavelength division multiplexing (DWDM) generated amounts of heat that could not be dissipated by some smaller package sizes. Further, smaller package sizes have a limited amount of space available for mounting and connecting additional components such as thermistors, monitor photodiodes, thermoelectric coolers, or impedance matching circuits.
Regarding smaller package sizes, it is desirable in fiber optic systems to use modular components so that a system can be created in a compact area and with as little expensive customization as possible. For example, many fiber optic systems are able to use modular transceiver modules. The modular transceiver modules include an input receiver optical subassembly (ROSA) and an output transmitter optical subassembly (TOSA). The ROSA comprises a photodiode for detecting optical signals and sensing circuitry for converting the optical signals to digital signals compatible with other network components. The TOSA comprises a laser for transmitting optical signals and control circuitry for modulating the laser according to an input digital data signal. The TOSA has an optical lens for collimating the light signals from the laser of the TOSA to an optical fiber. Additionally, the transceiver module includes pluggable receptacles for optically connecting the TOSA and the ROSA with other components within a fiber optic network.
The transceiver module often includes an electronic connector for connection to electrical components of the computer or communication device with which the transceiver module operates (a “host system”). The design of the transceiver, as well as other components within the fiber optic system, is standards-based, such that components can be connected without significant customization.
One particular pluggable standard that is currently being developed is the 10-Gigabit Small Form-factor Pluggable (XFP) standard. This standard defines various characteristics such as size, power consumption, connector configuration, etc. With regards to power consumption, the XFP standard references three power consumption levels of 1.5 W, 2.5 W and 3.5 W. When designing devices to operate within the XFP standard, attention must be given to what components are selected and how they are configured so as to not exceed the rated power consumption. These devices are constrained by principles of semiconductor physics to work preferentially in a certain temperature range. The module power dissipation and the package size and materials uniquely determine the module operating temperature for given ambient conditions, such as ambient temperature, airflow, etc. The resulting module operating temperature determines the types of optical and electronic components that can be successfully operated within the package. One such package is known as a transistor-outline header, otherwise known as a TO can or TO.
Transistor-outline headers are widely used in the field of opto-electronics, and may be employed in a variety of applications. As an example, transistor headers are sometimes used to protect sensitive electrical devices, and to electrically connect such devices to components such as printed circuit boards (“PCB”).
With respect to their construction, transistor headers often consist of a cylindrical metallic base with a number of conductive leads extending completely through, and generally perpendicular to, the base. With regard to the metallic base, the size of the base is often sized to fit within a specific TO standard size and lead configuration, examples of which include a TO-5 or TO-46. The leads are hermetically sealed in the base to provide mechanical and environmental protection for the components contained in the TO package, and to electrically isolate the conductive leads from the metallic material of the base. Typically, one of the conductive leads is a ground lead that may be electrically connected directly to the base.
Various types of devices are mounted on one side of the base of the header and connected to the leads. Generally, a cap is used to enclose the side of the base where such devices are mounted, so as to form a chamber that helps prevent contamination or damage to those device(s). The specific characteristics of the cap and header generally relate to the application and the particular device being mounted on the base of the header. By way of example, in applications where an optical device is required to be mounted on the header, the cap is at least partially transparent so to allow an optical signal generated by the optical device to be transmitted from the TO package. These optical TO packages are also known as window cans.
Although transistor headers have proven useful, typical configurations nevertheless pose a variety of unresolved problems. Some of such problems relate specifically to the physical configuration and disposition of the conductive leads in the header base. As an example, various factors combine to compromise the ability to precisely control the electrical impedance of the glass/metal feedthrough, that is, the physical bond between the conductive lead and the header base material. One such factor is that there are a relatively limited number of available choices with respect to the diameter of the conductive leads that are to be employed. Further, the range of dielectric values of the sealing glass typically employed in these configurations is relatively small. And, with respect to the disposition of the conductive leads, it has proven relatively difficult in some instances to control the position of the lead with respect to the through hole in the header base.
Yet other problems in the field concern those complex electrical and electronic devices that require many isolated electrical connections to function properly. Typically, attributes such as the size and shape of such devices and their subcomponents are sharply constrained by various form factors, other dimensional requirements, and space limitations within the device. Consistent with such form factors, dimensional requirements, and space limitations, the diameter of a typical header is relatively small and, correspondingly, the number of leads that can be disposed in the base of the header, sometimes referred to as the input/output (“I/O”) density, is relatively small as well.
Thus, while the diameter of the header base, and thus the I/O density, may be increased to the extent necessary to ensure conformance with the electrical connection requirements of the associated device, the increase in base diameter is sharply limited, if not foreclosed completely, by the form factors, dimensional requirements, and space limitations associated with the device wherein the transistor header is to be employed.
A related problem with many transistor headers concerns the implications that a relatively small number of conductive leads have with respect to the overall performance of the device and the need to connect additional circuitry required by certain types of laser when the transistor header is used. Semiconductor lasers circuits operate more efficiently when the circuit driving the semiconductor laser has an impedance that is equal to the impedance of the laser itself. There is a special need for impedance matching and load balancing when circuits are operating at relatively high frequencies, such as is the case in many semiconductor laser communication circuits. Mismatched circuits may cause transmission line reflections and a corresponding inability to maximize the power delivered to the semiconductor laser. Additionally, transmission line reflections can cause intensity noise and phase noise that results in transmission penalties in the fiber-optic circuit. Impedance matching is often accomplished through the use of additional electrical components such as resistors, capacitors, inductors, and transmission lines such as microstrips, striplines, or coplanar waveguides. However, such components cannot be employed unless a sufficient number of conductive leads are available in the transistor header. Thus, the limited number of conductive leads present in typical transistor headers has a direct negative effect on the performance of the semiconductor laser or other device.
In connection with the foregoing, another aspect of many transistor headers that forecloses the use of, for example, components required for impedance matching, is the relatively limited physical space available on standard headers. In particular, the relatively small amount of space on the base of the header imposes a practical limit on the number of components that may be mounted thereon. To overcome that limit, some or all of any additional components desired to be used must instead be mounted on the printed circuit board, some distance away from the laser or other device contained within the transistor header. Such arrangements are not without their shortcomings however, as the performance of active devices in the transistor header, such as lasers and integrated circuits, depends to some extent on the physical proximity of related electrical and electronic components. By minimizing the distance between the lasers and integrated circuits to the additional components required for impedance matching, the inherent transmission line between such components is minimized. As such, placing the components in close physical proximity reduces reflective transmission line losses.
Even when a sufficient number of contacts are available to connect external components to the laser for impedance matching, other problems arise. For example, one of the simplest methods of impedance matching is by shunting a resistive impedance across the laser source wherein the shunting impedance matches the impedance of the laser. The problem with this solution is that it adds an additional load to the power supply where the additional load is the shunt resistor and thus wastes power and generates heat.
In one example, suppose that a laser has a 25 ohm load impedance and a laser driver has a 12.5 ohm source impedance. To match the laser impedance, a 25 ohm resistor is shunted across the laser. This results in a 12.5 ohm load for the laser driver that, while impedance matched, requires more power to drive than if the laser driver only needed to drive a 25 ohm load. One way to eliminate the need for external components is to create an appropriately designed transmission line that transmits the laser signal from the laser driver to the laser itself, with proper characteristic impedance to match the laser and the laser driver. In this way, the laser driver efficiently supplies power to the 25 ohm load while minimizing harmful reflections. Such transmission lines are often appropriately sized microstrips, striplines, or coplanar waveguides, etc., formed on a printed circuit board using the characteristics of the conductive materials on the circuit board and the substrate on which the conductive materials are placed. As such, whereas transistor headers do not have internal printed circuit boards available, such matching transmissions lines cannot be constructed.
In addition to the need for matching circuits, there is also often a need for other additional circuitry. For example, an externally modulated laser (EML) comprises a laser and a semiconductor modulator. Examples of lasers that can be used with EMLs include a distributed feedback (DFB) laser or a distributed Bragg reflector (DBR) laser. Examples of modulators include an electroabsorptive modulator, in which the modulator absorbs light depending on a control voltage, or various interferometric modulators, such as the Mach-Zehnder modulator, often made with lithium niobate. An eternally modulated laser having an electroabsorptive modulator can be referred to as an EA EML (electroabsorbtive externally modulated laser). The integrated modulator has additional connections that require control signals from devices external to the transistor header that are normally not required when a laser without the integrated modulator is included. As such, without additional connections, lasers, such as EMLs, cannot be implemented in current transistor header designs.
The problems associated with various typical transistor headers are not, however, limited solely to geometric considerations and limitations. Yet other problems relate to the heat generated by components within, and external to, the transistor header. Specifically, transistor headers and their associated subcomponents may generate significant heat during operation. It is generally necessary to reliably and efficiently remove such heat to optimize performance and extend the useful life of the device.
However, transistor headers are often composed primarily of materials, Kovar® for example, that are not particularly good thermal conductors, but are instead selected for their properties of minimum thermal expansion and contraction, to match glass-metal seals and guarantee hermeticity. Such poor thermal conductivity does little to alleviate heat buildup problems in the transistor header components and may, in fact, exacerbate such problems. Various cooling techniques and devices have been employed in an effort to address this problem, but with only limited success. Such cooling problems have limited the types of lasers that may be used in transistor header applications. Particularly, such cooling problems have presented significant barriers to using lasers that are adapted for long-range fiber-optic communications such as externally modulated lasers (EMLs) that generate significant amounts of heat.
One drawback of using an EML is the heat that is generated by such a laser. Typically, most EMLs are operated between 25° C. to 30° C. As such, external cooling has commonly been required to pump heat away from the EML to maintain the laser at an appropriate operating temperature. The need for cooling components has previously imposed a limitation on the size of packages into which an EML is integrated. Further, because of the need for active cooling, the power consumption of a device integrating an EML is often greater than that allowed by many of the smaller package size standards such as XFP. Previously, EMLs have not been effectively integrated into smaller packages because of these cooling requirements. Additionally, in order to keep a laser's wavelength stable to enable such applications as DWDM, the temperature must be finely controlled to be fixed regardless of varying ambient temperatures and conditions. One of the best methods to accomplish this temperature control is to have precise control of the same cooler that is used to keep the laser at an appropriate operating temperature.
Solid state heat exchangers may be used to remove some heat from transistor header components. However, the effectiveness of such heat exchangers is typically compromised because, due to variables such as their configuration and/or physical location relative to the primary component(s) to be cooled, such heat exchangers frequently experience a passive heat load that is imposed by secondary components or transistor header structures not generally intended to be cooled by the heat exchanger. The imposition on the heat exchanger of such passive heat loads thus decreases the amount of heat the heat exchanger can effectively remove from the primary component that is desired to be cooled, thereby compromising the performance of the primary component.
As suggested above, the physical location of the heat exchanger or other cooling device has various implications with respect to the performance of the components employed present in the transistor header. One particular problem in the context of thermoelectric cooler (“TEC”) type heat exchangers arises because TECs have hot and cold junctions. The cold junction, in particular, can cause condensation if the TEC is located in a sufficiently humid environment. Such condensation may materially impair the operation of components in the transistor header, and elsewhere.
Solid state coolers, such as TECs, are intrinsically very inefficient devices. State-of-the-art coolers have efficiencies measured in single or low double digits. Thus, the power consumption becomes astronomical when an attempt is made to cool lasers in packages that have significant thermal leaks. This process requires large amounts of power, which is inappropriate for small devices because it causes large temperature rises and because it is not permitted under standards, such as the XFP standard.
Another concern with respect to heat exchangers is that the dimensions of typical transistor headers are, as noted earlier, constrained by various factors. Thus, while the passive heat load placed on a heat exchanger could be at least partly offset through the use of a relatively larger heat exchanger, the diametric and other constraints imposed on transistor headers by form factor requirements and other considerations place practical limits on the maximum size of the heat exchanger.
Finally, even if a relatively large heat exchanger could be employed in an attempt to offset the effects of passive heat loads, large heat exchangers present problems in cases where the heat exchanger, such as a TEC, is used to modify the performance of transistor header components such as lasers. For example, by virtue of their relatively large thermal mass or load, such heat exchangers are not well suited to implementing the rapid changes in laser performance that are required in many applications, because such large heat exchangers cannot transfer the heat rapidly enough. Moreover, the performance of the laser or other component may be further compromised if the heat exchanger is located relatively far away from the laser because the thermal resistance is proportional to the distance between the component and the heat exchanger.
In light of the above discussion, a need exists for a transistor header design for use within an optical transceiver module that overcomes the above challenges. In particular, a transistor header is needed that enables a relatively greater number of interconnects to be established between the header interior and devices located outside of the header. Such a solution should also enable the use of components heretofore unimplemented in current header designs, such as EML laser configurations and thermo-electric coolers.